Canon 5DS vs 7D2 (image quality not purchase choice)

[Brachytron]

With all the excitement about the high resolution of the new 5DS, I am keen to understand what is so much better than the 7D2 and if so, why. The pixel dimensions of the two cameras are very nearly the same (the 5DS in 1.6 crop mode is ~19MP and the 7D2 is ~20MP) with a slight size advantage for the 5DS but perhaps not enough to warrant the extra resolution claims. On this basis, I would expect the two cameras at base ISO to have the same or at least very similar resolution. Also, the dynamic range of the 5DS is higher at 12.4 vs 11.8 on the 7D2. Again, with the same pixel size why is this the case? I can't see that a larger sensor size can affect these because both measures are related to pixel size (all other things being equal of course). I am sure there are some subtle tweeks to the 5DS that aren't on the 7D2, but the I guess the essential sensor tech is similar because they were released close together.

I can understand why lenses perform better on FF, such as the 1DX vs the 7D with both at 18MP, if they aren't able to resolve to the smaller pixel resolution of the APS-C camera, but when the pixel resolution of a FF camera matches that of an APS-C, then I would expect lenses to perform the same. Is this the case?

Am I missing something? If so I'd be glad to enter a debate. Thanks.

 

[pingu666]

I think it comes down to 50mp in one shot, and I suspect the 5dsr has more sharpening to its files out of the box vs the competition...

you could get similar end result by stiching a crop camera, but that's tedious...

 

[HoppyUK]

<snip>
Am I missing something? If so I'd be glad to enter a debate. Thanks.

Yes, two fundamentals - both concerning the physical size (area) or the sensors.

The FF sensor has more than twice the image area. It collects double the light, hence the improvement in dynamic range and, in theory at least, ISO performance. The other thing is sharpness, that has much more to do with physical sensor size than the number of pixels. Fact of physics is when lens resolution demands go up (smaller sensor) image contrast goes down - and it's contrast than contributes most to perceived sharpness (basic MTF theory).

Bottom line is, all things being equal (as they mostly are with similar generation cameras) FF always delivers better image quality than physically smaller sensors.

 

[Brachytron]

It collects double the light, hence the improvement in dynamic range and, in theory at least, ISO performance..

I think that's part of my lack of understanding. Why do pixels outside the sensor area of an APS-C sensor (i.e. in the area between FF sensor and APS-C sensor) create more dynamic range? Isn't the dynamic range a function of the pixels themselves rather than the image area?

Fact of physics is when lens resolution demands go up (smaller sensor) image contrast goes down - and it's contrast than contributes most to perceived sharpness (basic MTF theory)...

Is that smaller sensor or smaller pixels? I get that when the pixel size is larger than the Airy disc created by the lens, then the sensor is the limit to resolution, however when the Airy disc is larger than the pixel size, the lens is the limiting factor, contrast is lowered and MTF reduces. But I don't get what the actual sensor size has to do with it. It still looks to be pixel size related.

 

[HoppyUK]

I think that's part of my lack of understanding. Why do pixels outside the sensor area of an APS-C sensor (i.e. in the area between FF sensor and APS-C sensor) create more dynamic range? Isn't the dynamic range a function of the pixels themselves rather than the image area?

Dynamic range and noise (ISO) is directly related to how much light is collected (photons) and how much the signal needs to be amplified to create an image. Bigger sensor collects more light.

Is that smaller sensor or smaller pixels? I get that when the pixel size is larger than the Airy disc created by the lens, then the sensor is the limit to resolution, however when the Airy disc is larger than the pixel size, the lens is the limiting factor, contrast is lowered and MTF reduces. But I don't get what the actual sensor size has to do with it. It still looks to be pixel size related.

It's smaller sensor, with smaller physical size/area, that reduces sharpness. Ignore the stuff about Airy disc size and pixel/resolution limits. It's a theory, and only half a theory at that - it ignores contrast that is more significant than resolution in our perception of sharpness (pretty much regardless of pixel count, within reason). It's simply this - smaller sensors demand greater resolution from the lens, in line with the crop factor, because they have to be enlarged more for a given size output/print. Eg, for the same standard of sharpness, the lens on an APS-C 1.6x camera has to deliver the same level of contrast at, say, 38-lines-per-mm whereas on full-frame it only needs to work at 24-lpmm (24 x 1.6 = 38). That's too big a gap for even the best lenses to bridge, therefore contrast is higher on FF and the image looks sharper.

 

[Chipper]

I watched a presentation at a Canon roadshow by David Ward... The file size is enormous..and very small crops gave amazing results. 270mb for a TIF.. Make sure you have the hardware to back it up if you treat yourself to one in the future!

 

[seriousrikk]

I think that's part of my lack of understanding. Why do pixels outside the sensor area of an APS-C sensor (i.e. in the area between FF sensor and APS-C sensor) create more dynamic range? Isn't the dynamic range a function of the pixels themselves rather than the image area?

Dynamic range measurements are done on the files output from the sensor.

So until someone measures dynamic range of the files output in the crop mode then I don't see how you can really compare it.

 

[Brachytron]

It's simply this - smaller sensors demand greater resolution from the lens, in line with the crop factor, because they have to be enlarged more for a given size output/print. Eg, for the same standard of sharpness, the lens on an APS-C 1.6x camera has to deliver the same level of contrast at, say, 38-lines-per-mm whereas on full-frame it only needs to work at 24-lpmm (24 x 1.6 = 38). That's too big a gap for even the best lenses to bridge, therefore contrast is higher on FF and the image looks sharper

I get it, - enlargement, thanks!. So, if the 5DS were in 1.6 crop mode, or if the image from the 5DS were made the same size as the 7D2 (e.g. a bird in the shot, not pixel dimensions), the resolution would very nearly a match because you are then enlarging the pixels in the 5DS?

Still don't get why a larger sensor generates more dynamic range. I understand why larger pixels in a FF camera generate large DN, because they collect more light and the signal is amplified less, however whether FF or APS-C with the same pixel size the light gathering power of the pixel is the same, as is the amplification, so therefore dynamic range should be the same too.

Thanks for the discussion so far.

 

[Brachytron]

I watched a presentation at a Canon roadshow by David Ward... The file size is enormous..and very small crops gave amazing results. 270mb for a TIF.. Make sure you have the hardware to back it up if you treat yourself to one in the future!

I do find it amusing when photo mags review cameras and rave about the high resolution files one minute but then criticise the file size the next.

 

[Canon Bob]

Still don't get why a larger sensor generates more dynamic range.

This explanation is based on 8 years old information and may still apply.

When you enlarge a captured image then noise becomes more prevalent and APS-C files need more enlargement to achieve the same final output size. Historically (and possibly today) manufacturers clipped the DR to reduce shadow noise...ie, a choice between noise or a larger DR. Back in 2008 Nikon produced the cleaner files and Canon had a slightly better DR (all things being equal). Nikon achieved the cleaner files by clipping the bottom of the DR to remove the noise whilst Canon left it there to provide maximum (acceptable) DR.

I suspect it's the same today and Canon have a little more latitude on their FF sensor (given the same pixel density).

Bob

 

[HoppyUK]

I get it, - enlargement, thanks!. So, if the 5DS were in 1.6 crop mode, or if the image from the 5DS were made the same size as the 7D2 (e.g. a bird in the shot, not pixel dimensions), the resolution would very nearly a match because you are then enlarging the pixels in the 5DS?

Still don't get why a larger sensor generates more dynamic range. I understand why larger pixels in a FF camera generate large DN, because they collect more light and the signal is amplified less, however whether FF or APS-C with the same pixel size the light gathering power of the pixel is the same, as is the amplification, so therefore dynamic range should be the same too.

Thanks for the discussion so far.

It's better to ignore pixels for the whole of this debate. Pixel density etc is not irrelevant, but very distracting and leads away from the main point. It's light, total photon capture that is the main driver, whether photons are collected in one larger pixel or two smaller ones.

Edit: 'equivalence' is at the heart of this - good article here http://www.dpreview.com/articles/2666934640/what-is-equivalence-and-why-should-i-care